复杂异质分布下的雷达目标检测技术研究

发布时间：2017-12-28 23:02

本文关键词：复杂异质分布下的雷达目标检测技术研究　出处：《哈尔滨工业大学》2016年硕士论文　论文类型：学位论文

【摘要】：复杂异质的杂波背景是目前机载非正侧视雷达目标检测技术研究面临的主要问题。非正侧视的工作方式使地杂波回波产生距离与多普勒频率间的依赖性,近程杂波在杂波背景中形成一条明显的眉毛线。这种情况对传统的空时自适应处理(Space-Time Adaptive Processing,STAP)算法造成较大影响,异质样本使协方差矩阵估计精度下降。本文针对这种复杂异质的杂波背景,提出利用样本的统计特性进行同质性检验,旨在将统计特性相同的样本数据用同一分区标识进行标记作为STAP算法的前期处理。接着在进行常用的STAP处理时,根据不同的分区标识进行协方差矩阵估计样本的选择,提高估计精度,改善STAP算法应用到实际雷达数据的性能。本文的具体研究内容如下:1、对机载雷达回波数据特性的分析,及传统STAP方法原理的介绍。分析了机载雷达由于正前视的工作方式,带来了杂波的距离多普勒依赖性,导致杂波背景复杂异质。同时介绍传统STAP算法的原理,STAP杂波抑制性能取决于协方差矩阵的估计精度,用于估计的样本要求独立同分布。此部分研究主要为后续根据样本统计特性进行数据分区奠定理论基础。2、对同质性检验方法进行深入研究。对E距离同质性检验、AD同质性检验、K样本AD检验(K-AD同质性检验)的原理进行了介绍。通过仿真实验验证它们具有分辨不同概率分布的能力,同时对各自的性能进行探究。经过对实验结果及运行时间的比较,本文决定采用K样本AD检验作为本文所使用的同质性检验方法,记做K-AD同质性检验。3、提出基于K-AD同质性检验的分区算法。K-AD同质性检验是通过计算两序列间的K-AD检测量,并对比检测量与经验阈值间的大小关系进行判断两序列是否满足相同概率分布。本文根据实际雷达回波数据的形式,利用参考窗口制定适用的分区算法,得到综合考虑到整体数据样本统计特性的分区结果。4、提出基于K-AD同质性检验分区的STAP算法。基于K-AD同质性检验的分区算法能够实现对实际雷达回波数据中统计特性相同的数据样本进行划分,根据此分区则可以选择与检测样本相同概率分布的数据样本作为参考单元进行有效的协方差矩阵估计。本方法的重点在于参考单元的选择,后续的STAP算法采用的是性能较优越的3DT-STAP。采用在仿真数据及实测数据中加入仿真目标的方式,观察目标处处理前后的信干噪比(SINR)变化,验证该算法具有杂波抑制的能力,能够实现对杂波的有效抑制。5、利用广义内积算法、自适应功率剩余算法、功率选择训练算法及基于K-AD同质性检验分区的STAP算法对实际雷达数据进行处理。通过考察处理后仿真目标位置处的SINR大小分析算法处理性能,对比观察本文所提算法的杂波抑制能力。
[Abstract]:Complex heteroclutter background is the main problem in the research of airborne non - positive side view radar target detection technology. The way of non side looking work makes the dependence of the distance between the clutter echo and the Doppler frequency on the ground clutter, and the short distance clutter forms a distinct eyebrow line in the clutter background. This situation has great influence on the traditional Space-Time Adaptive Processing (STAP) algorithm, and the accuracy of covariance matrix estimation is reduced by heterogeneous samples. Aiming at this complex and heterogeneous clutter background, we propose to use the statistical characteristics of samples to homogeneity test, aiming to mark the same sample data with the same partition mark as the pre processing of STAP algorithm. Then, in the commonly used STAP processing, the selection of the covariance matrix estimation samples is done according to the different partition identities, which improves the estimation accuracy and improves the performance of the STAP algorithm applied to the actual radar data. The main contents of this paper are as follows: 1. The analysis of the characteristics of the airborne radar echo data and the introduction of the principle of the traditional STAP method. This paper analyzes the working mode of airborne radar which is forward looking forward, which brings the distance Doppler dependence of clutter, which leads to complex background heterogeneity of clutter. At the same time, the principle of traditional STAP algorithm is introduced. The performance of STAP clutter suppression depends on the estimation accuracy of covariance matrix, and the estimated samples require independent and identical distribution. This part of the study lays a theoretical foundation for subsequent data zoning according to the statistical characteristics of the sample. 2. Study on the method of homogeneity test. The principle of E distance homogeneity test, AD homogeneity test and K sample AD test (K-AD homogeneity test) is introduced. The simulation experiments show that they have the ability to distinguish the different probability distribution, and explore their respective performance. After comparing the experimental results and the running time, this paper decides to use the K sample AD test as the homogeneity test method used in this paper, and record the K-AD homogeneity test. 3. A partition algorithm based on K-AD homogeneity test is proposed. K-AD homogeneity test is to determine whether the two sequences satisfy the same probability distribution by calculating the K-AD detection volume between two sequences, and comparing the relationship between the detection volume and the empirical threshold. Based on the actual radar echo data, we use the reference window to draw up the applicable partition algorithm, and get the partition result considering the statistical characteristics of the whole data sample. 4, a STAP algorithm based on K-AD homogeneity test partition is proposed. The K-AD algorithm partition homogeneity test can be achieved on the statistical characteristics of the radar echo data in the same data samples based on the partition, the partition can choose according to the same sample and detection probability distribution of the data sample as a reference unit for covariance matrix estimation effectively. The focus of this method is the selection of the reference unit, and the subsequent STAP algorithm uses a better performance 3DT-STAP. Using the way of adding simulation target in simulated data and measured data, we can observe the change of signal to noise ratio (SINR) before and after processing, verify that the algorithm has the ability of clutter suppression, and can effectively suppress clutter. 5, we use generalized inner product algorithm, adaptive power residual algorithm, power selection training algorithm and STAP algorithm based on K-AD homogeneity test partition to process actual radar data. By investigating the performance of the SINR size analysis algorithm at the position of the simulated target, the clutter suppression ability of the proposed algorithm is compared.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN957.51

【相似文献】